Getting Norwegian meta‑description text during conversation transcription

> I recorded a conversation in Turkish in the web demo (Edge on Windows). Instead of a Turkish transcript, I get this Norwegian meta‑description:
> “Kort, svært utydelig ytring på et skandinavisk språk (trolig norsk/svensk dialekt) med mange feiltolkede ord. Ingen klare temaer, planer eller konkrete avtaler kan tryggt identifiseres.”
> My system/browser languages are Turkish/English only, no Scandinavian languages. It looks like my Turkish audio is being misrouted to a Scandinavian ASR/classifier that then outputs this generic description. I need the ability to explicitly select Turkish as the transcription language and avoid this Scandinavian model routing.